8960 Biochemistry 1995,34, 8960-8972

X-ray Structures of the Myosin Motor Domain of Dictyostelium discoideum Complexed with MgADPBeF, and MgADPAlF4-

Andrew J. Fisher,@ Clyde A. Smith,@ James B. Thoden,@ Robert Smith,@ Kazuo Sutoh," Hazel M. Holden,@ and Ivan Rayment*.@

Institute for Enzyme Research and Department of Biochemistry, University of Wisconsin, Madison, Wisconsin 53705, and Department of Pure and Applied Sciences, University of Tokyo, Komba, Tokyo 153, Japan

Received March 15, 1995; Revised Manuscript Received May 3 , 1995@

ABSTRACT: The three-dimensional structures of the truncated myosin head from Dictyostelium discoideum myosin I1 complexed with beryllium and aluminum fluoride and magnesium ADP are reported at 2.0 and 2.6 A resolution, respectively. Crystals of the beryllium fluoride-M ADP complex belong to space group P21212 with unit cell parameters of a = 105.3 A, b = 182.6 1, and c = 54.7 A, whereas the crystals of the aluminum fluoride com lex belong to the orthorhombic space group C2221 with unit cell

obtain these crystals. These structures reveal the location of the nucleotide complexes and define the amino acid residues that form the active site. The tertiary structure of the protein complexed with MgADPBeF, is essentially identical to that observed previously in the three-dimensional model of chicken skeletal muscle myosin subfragment-1 in which no nucleotide was present. By contrast, the complex with MgADPeAlF4- exhibits significant domain movements. The structures suggest that the MgADPBeF, complex mimics the ATP bound state and the MgADPoAlF4- complex is an analog of the transition state for hydrolysis. The domain movements observed in the MgADPAlF4- complex indicate that myosin undergoes a conformational change during hydrolysis that is not associated with the nucleotide binding pocket but rather occurs in the COOH-terminal segment of the myosin motor domain.

dimensions of a = 87.9 A, b = 149.0 !i , and c = 153.8 A. Chemical modification was not necessary to

The molecular mechanism by which chemical energy is transduced into directed movement is a fundamental question in biology. It is well established that muscle contraction occurs when the thick and thin filaments, which are built primarily of myosin and actin, slide past each other (Huxley & Niedergerke, 1954; Huxley & Hanson, 1954). Movement occurs through the interaction of the globular head of myosin with actin and is driven by the hydrolysis of ATP.

Recently, the three-dimensional structures of both actin and myosin were determined by X-ray crystallography (Holmes et al., 1990; Kabsch et al., 1990; Rayment et al., 1993b), and combination of these results with those from electron microscopy gave rise to a structural hypothesis for how myosin converts chemical energy into directed move- ment (Milligan & Flicker, 1987; Milligan et al., 1990; Rayment et al., 1993a). The major tenet of this hypothesis was that domain movements in myosin, induced by the binding of ATP to the actin-bound state, generate a cyclical series of conformational changes that reduce the affinity of myosin for actin and place the myosin in a state in which it

+This work was supported by NIH Grant AR35186 to I.R. C.A.S. is a Fogarty International Research Fellow (TW05194). A.J.F. and J.T. are supported by NSRA Fellowships -08304 and GM15950, respec- tively. K.S. is supported by a grant-in-aid from the Minister of Education, Science and Culture (06404081) and the Mitsubishi Founda- tion and the Ueha Memorial Foundation.

t The coordinates have been deposited in the Brookhaven Protein Data Bank (file names lMMD and IMND).

* Address correspondence to this author at the Institute for Enzyme Research, University of Wisconsin, 1710 University Ave., Madison, WI 53705.

4 University of Wisconsin. II University of Tokyo. @Abstract published in Advance ACS Abstracts, July 1 , 1995.

can hydrolyze the nucleotide. Movement of myosin relative to actin was suggested to occur through reversal of the induced conformational changes as myosin rebinds to actin and releases the reaction products. This model was consis- tent with previous kinetic studies, both in solution and in fibers, which have established that the power stroke, or energy transduction process, occurs during product release (Goldman, 1987; Lymn & Taylor, 1971). Central to this hypothesis was the suggestion that myosin interacts with actin in a stereospecific manner such that the onset of binding requires the molecules to be in a unique orientation before the power stroke is initiated. This hypothesis left questions unanswered concerning the nature of the structural transitions that underlie this process and the mechanism of ATP hydrolysis.

One of the most important questions raised by the initial model of myosin subfragment-1 (myosin S1)' was what is the structure of the myosin head with ATP bound? It was suggested that the nucleotide binding pocket closes when ATP binds to myosin and generates a large movement of the COOH terminus of the myosin head relative to the actin binding conformation (Rayment et al., 1993a). This change would then prepare the molecule to unbend as it reattached to actin and to release the products of hydrolysis. Given the complexity of the myosin head, it was impossible to

I Abbreviations: myosin S 1 , myosin subfragment-I; S lDc, Dicryo- stelium discoideum myosin I1 motor domain; PMSF, phenylmethane- sulfonic acid; S 1 DcBeF,.MgADP, the beryllium fluoride-ADP com- plex of D . discoideum myosin head; SlDcAIF4*MgADP, the aluminum fluoride-ADP complex of D. discoideum myosin head; BTP, Bis-Tris- propane; DT", dithiothreitol; HEPES, N-(2-hydroxyethyl)piperazine- K-2-ethanesulfonic acid; rms, root mean square; Pi, phosphate; EDTA, ethylenediaminetetraacetic acid.

0006-2960/95/0434-8960$09.00/0 , -. , 0 1995 American Chemical Society

Structural Studies on Myosin-Nucleotide Complexes

predict or model accurately the nature of these conforma- tional changes.

Another question arose from the strategy adopted to obtain crystals of chicken skeletal myosin S1 (Rayment et al., 1993b). In that study it was found necessary to reductively methylate the protein in order to grow crystals suitable for structural analysis (Lundblad, 1991; Means & Feeney, 1968). Myosin retains its enzymatic activity after this modification even though many of the rate constants associated with the steps in the hydrolytic process are altered (White & Rayment, 1993). Control studies, in which the structure of hen egg white lysozyme was determined after chemical modification, revealed that reductive methylation had minimal effect on its tertiary structure (Rypniewski et al., 1993). It remained to be determined, however, whether the reductive alkylation had significantly altered the relative arrangement of domains within myosin. Recently, the use of the X-ray structure as a model for the contractile cycle has been questioned due to the changes in the enzymatic properties of the molecule induced by reductive methylation (Bivin et al., 1994; Phan et al., 1994).

Questions conceming the structure of the myosin head also remained unanswered because the original structure deter- mination was only at 2.8 8, resolution. Although this revealed the path of the polypeptide chain, it was not of high enough resolution to categorically define the geometry of all of the important residues in the active site or the water structure. Identification of the hydrogen-bonding patterns and the water structure within the active site is critical, since hydrolysis occurs by attack of a water molecule on the y-phosphorus atom of ATP. The restricted quality of the diffraction from the crystals arose not only from the numerous myosin heavy and light chain isoforms and from the proteolytic heterogeneity introduced in their preparation but also from its asymmetric shape.

The questions outlined above can be addressed only by determination of the structure of the myosin head in presence of an ATP analog, without chemical modification, and from a crystal form that diffracts to higher resolution. Chemical studies have shown that it is possible to trap ADP in the active site of myosin in a variety of ways including cross- linking the two reactive sulfydryl groups in the myosin heavy chain and through the formation of long-lived complexes of ADP with phosphate analogs. These studies have proved to be of enormous value in dissecting the catalytic cycle and defining the environment of the active site (Yount et al., 1992). Phosphate analogs include vanadate (Goodno, 1982) and more recently beryllium fluoride and aluminum fluoride (Phan & Reisler, 1992; Phan et al., 1993; Werber et al., 1992). It has been suggested that these complexes are analogs of the metastable ADP,Pi state that occurs after hydrolysis (Maruta et al., 1993; Phan et al., 1993). The beryllium and aluminum fluoride complexes are of particular interest since they are very stable and thus are good candidates for structural studies (Maruta et al., 1993). We report here the structures of the MgADP complexes of the motor segment of myosin I1 from Dictyostelium discoideum in the presence of beryllium and aluminum fluoride at 2.0 and 2.6 8, resolution, respectively. These structures suggest that the MgADPBeF, complex mimics the ATP bound state and the MgADPAlF4- complex is an analog of the transition state for hydrolysis. Myosin from Dictyostelium discoideum

Biochemistry, Vol. 34, No. 28,1995 8961

was chosen for this study since an expression system for chicken skeletal muscle myosin was not available.

EXPERIMENTAL PROCEDURES

Protein Design and Purification. The myosin head is a highly asymmetric molecular assembly (Rayment et al., 1993b). It is characterized by a cluster of domains that define a motor unit, which is attached to an extended motif where the myosin light chains bind. The motor unit in skeletal muscle myosin is built from approximately 780 amino acid residues. The light chain binding motif, also called the regulatory domain (Xie et al., 1994), is built around an 85 8, long a-helix that leaves the motor unit and extends throughout the remainder of the myosin head. The three- dimensional structure of the isolated light chain binding motif in scallop has been determined independently and confirms that observed in chicken skeletal myosin S1 (Xie et al., 1994). Examination of the myosin head suggested it might be possible to generate a shorter head that contained all of the elements of the molecular motor and might be more suitable for crystallization. This has been accomplished by expres- sion of a truncated head from the myosin I1 of D . discoideum (Itakura et al., 1993) with the system developed by Spudich and co-workers (Manstein et al., 1989). This truncated head retains its ATPase activity although the kinetic rate constants have changed (Sally Woodward, personal communication). It also exhibits a limited ability to move actin, suggesting that it contains the fundamental unit necessary to generate movement (Itakura et al., 1993).

The myosin heavy chain was truncated genetically at residue Ile 762 to yield a fragment with molecular weight -87 kDa. The site of truncation was equivalent to residue Leu 783 in chicken skeletal muscle myosin. The fragment studied here consists of residues Asp 2 to Asn 762 in the sequence of Dictyostelium myosin (Warrick et al., 1986). It is presumed that the initial methionine has been removed since no density is observed for this residue. In the original sequence the second amino acid was an asparagine residue. This was converted to an aspartate during the construction of the fragment. The side chain of Asp 2 forms a salt bridge to Arg 146 in a manner that is consistent with this amino acid. Additionally, the final three residues were changed from Gln-Arg-Ile in the original sequence to Leu-Pro-Asn to create a restriction site (Itakura et al., 1993). The motor unit of Dictyostelium myosin is smaller than that of chicken skeletal muscle myosin due to shorter loops that connect the major tryptic fragments (Balint et al., 1975; Momet et al., 1979; Warrick et al., 1986).

The truncated myosin head from D . discoideum (SlDc) was prepared as before with modifications (Itakura et al., 1993). The Dictyostelium cells were grown in axenic culture (HL5 medium) and harvested at saturation (1 -2 x lo7 cells/ mL) by centrifugation. The cells were washed in 10 mM Tris, pH 8, and then frozen in liquid Nz. The isolation of the protein was carried out at 4 "C. Approximately 50 g of cells was resuspended by sonication in 400 mL of 20 mM Bis-Tris-propane (BTP), pH 7.5,25 mM NaCl, 4 mM EDTA, 1 mM DTT, and 0.2 mM phenylmethanesulfonyl fluoride (PMSF). The cellular debris was removed by centrifugation at lOOOg for 10 min, whereupon the cytoskeletal components were collected by a high-speed centrifugation at 235 OOOg for 1 h. The myosin was extracted from the pellets by the

8962 Biochemistry, Vol. 34, No. 28,1995 Fisher et al.

Table 1: X-ray Data Collection Statistics complex source and detector 1 T ("C) total unique redundancy completion (%) resolution (A) R,,,O

MgADPBeFx*S 1Dc Siemens HI-STAR 1.542 4 32930 2.5 98 2.6 11.0 MgADPBeF,-SlDc SSIU, MAR scanner 1.08 4 63111 6.3 88 2.0 3.3 MgADPAIF4SlDc Siemens 1.542 -160 29506 2.3 96 2.6 6.1

Rmcrge = ={I - I I}/&h x 100, where [hi and l h are the intensities of the individual and mean st~cture factors.

addition of 20 mM BTP at pH 7.5, 50 mM NaCl, 7 mM MgCl2,6 mM ATP, 1 mM DTT, and 0.2 mM PMSF. The actin and other cytoskeletal components were removed by centrifugation at 250 OOOg for 80 min. The supematant was applied to a 50 mL Toyopearl DEAE-650M column to remove any residual actin. SlDc was precipitated by the addition of solid ammonium sulfate to 70% saturation. The precipitate was recovered by centrifugation and dialyzed against 50 mM Tris-HC1, pH 8.0, containing 100 mM KCl, 2 mM MgC12, 0.2 mM NaN3, 1 mM DTT, and 0.2 mM PMSF. The resultant solution was clarified by centrifugation and applied to an ATP affinity column where the attachment of the nucleotide to the column substrate was via the ribose 2'- or 3'-hydroxyl group (synthesized by Ralph Yount, Washington State University). S 1Dc was eluted with 50 mM Tris-HC1 at pH 8.0 containing 100 mM KC1, 2 mM sodium pyrophosphate, 2 mM EDTA, 0.2 mM NaN3, and 1 mM DTT. The protein was concentrated to 8-10 mg/mL by microfiltration and exchanged simultaneously into 10 mM HEPES, pH 8.0, 100 mM NaC1, 2 mM MgC12, 0.2 mM NaN3, and 1 mM DTT. A typical preparation yields 3 mg of protein from 50 g of cells.

Crystal Growth. Crystals of the beryllium fluoride- magnesium ADP complex with SlDc were grown by microbatch from a mixture of final composition 4-5 mg/ mL protein, 8.5% poly(ethy1ene glycol) 8000, 25 mM HEPES pH 7.0, 1.5% methylpentanediol, 170 mM NaCl, 1.8 mM ADP, 1.8 mM MgC12, 1.8 mM BeC12,9 mM NaF, and 3 mM DTT at 4 "C. The crystals of the aluminum fluoride complex were grown under similar conditions except that the beryllium chloride was replaced by 1.75 mM Al(N03)3 and the sodium fluoride concentration lowered to 6 mM. Crystallization was induced by the introduction of either micro- or macroseeds obtained from preliminary hanging- drop experiments. The protein was not chemically modified.

X-ray Data Collection and Processing. Crystals of the beryllium fluoride-MgADP complex (MgADPBeF,-S 1 D c ) ~ belong to space group P21212 with unit cell parameters of a = 105.3 A, b = 182.6 A, and c = 54.7 A. There is one molecule per asymmetric unit (VM = 3.0 A3/Da, solvent content 59%). These crystals show diffraction to 1.8 A resolution with synchrotron radiation, although data to 2.0 8, resolution were utilized in the present study. An initial X-ray data set was recorded on a Siemens HI-STAR double- detector system at 4 "C and extended to higher resolution at the Stanford Synchrotron facility (SSRL) with a MAR image plate detector. In contrast, crystals of the aluminum fluoride complex (MgADPAlF4.S 1 Dc) belong to the orthorhombic space group C2221 with unit cell dimensions of a = 87.9 A, b = 149.0 A, and c = 153.8 A. The crystals contain one molecule per asymmetric unit (VM = 2.9 A3/Da, solvent ~~ ~

The exact composition of the beryllium fluoride species is unknown (Henry et al., 1993; Maruta et al., 1993). Consequently, these complexes will be designated as BeF, and AIF4-.

content rz 58%). X-ray data were collected from a single crystal at - 160 "C on a Siemens HI-STAR area detector. A summary of the data collection statistics is reported for both crystal forms in Table 1.

Structure Determination of MgADPBeF,*SIDc. The struc- ture of MgADPBeF,*SlDc was solved by molecular replace- ment from a phasing model based on the chicken myosin SI structure composed of residues Glu 74 to Glu 776 (Rossmann, 1972). This excluded the NH2-terminal /?-barrel and the light chain binding motif. All the nonconserved residues were truncated to alanines in the chicken S 1 search model. Area detector data between 9.0 and 4.5 A resolution were used in a Patterson correlation refinement of the 51 strongest rotation peaks (Briinger, 1990a,b) and generated a single peak which was 2.3~7 greater than the next highest peak. The highest peak in the translation search (1.3~7 greater than the next highest) gave in an R-factor of 49.3% for all data to 3.5 A resolution. Identical solutions were obtained with the molecular replacement programs GLRF and AMORE (Navaza, 1993; Tong & Rossmann, 1990). The phasing model was divided into five components that were allowed to vary independently during rigid body refinement. This decreased the R-factor to 44.7%. An initial electron density map was computed using SIGMAA coefficients to suppress the model bias (Read, 1986) and revealed density for most of the side chains that were excluded from the phasing model as well as strong density for the MgADPBeF, complex. In the initial map most of the residues between Asp 75 and Ile 696 were visible. A few strands were observed for the NH2- terminal /?-barrel domain. The last residue observed at the carboxyl-terminal end was Ile 696. After the first round of model building, the structure was refined by the simulated annealing algorithm in X-PLOR (Briinger, 1990b) followed by conventional least squares refinement with the program TNT (Tronrud et al., 1987). This reduced the R-factor to 35.0% for all recorded data to 2.7 A resolution. Thereafter, the model was refined against the synchrotron data (SSRL) to 2.0 A resolution. After refinement the final model included 743 residues together with 375 water molecules and the nucleotide complex. The final R-factor between the observed and calculated data was 19.1 % for all data recorded between 30 and 2.0 A resolution where the rms deviation from ideality for the bond lengths, angles, and trigonal planes was 0.015 A, 2.26", and 0.007 A, respectively. A section of representative electron density for the protein and the nucleotide complex in the active site pocket is shown in parts a and b of Figure 1, respectively.

Structure Determination of MgADPAlFsSIDc. The alu- minum fluoride complex was solved by molecular replace- ment (Rossmann, 1972) from the structure of MgADP BeF,*SlDc and refined in a similar manner. Data between 10.0 and 4.0 A were included in the rotation and translation searches. A Patterson correlation refinement of the overall orientation and of the individual orientations of five frag- ments of the S1 model was performed on the top 200 peaks

Structural Studies on Myosin-Nucleotide Complexes Biochemistry, Vol. 34, No. 28, I995 8963

b

C

FIGURE 1: (a) Representative electron density for a segment including Lys 130 near the active site pocket in the MgADPBeF,*SlDc complex at 2.0 8, resolution. (b) Electron density for the nucleotide complex in the active site pocket of the MgADPBeFx*SIDc complex. (c) Electron density for the MgADP*AlF& 1Dc complex. In (b) and (c) aluminum, beryllium, carbon, fluorine, magnesium, nitrogen, oxygen, and phosphorus are depicted in cyan, green, black, yellow, orange, blue, red, and pink, respectively. The majority of the electron density in the central segment of the myosin heavy chain is of this quality in the MgADP*BeFx*SIDc complex. Although the protruding NH2- terminal domain and the COOH-terminal domain in the MgADP-BeFx-S 1Dc structure have higher temperature factors, the chain tracing is unequivocal. Although the density for the AlF4- complex appears spherical, when these atoms are excluded from the phase calculation the difference density is clearly square planar, providing definitive positions for the fluorine ligands. This figure was prepared from a plot file generated from the molecular graphics program FRODO (Jones, 1985) and drawn with the program MOLVIEW (Smith, 1993). A cover radius of 2.0 8, was chosen to exclude density not associated with the atoms shown.

from the rotation search. Several strong solutions emerged almost all of the non-glycinal residues lie within the fully from the Patterson correlation refinement, all of which were allowed regions o e Ramachandran plot (Figure 2). The close to the highest peak. A translation search with this few exceptions li er in loops or at crystal contacts. orientation yielded a peak 60 greater than the next highest, with an R-factor of 48.7%. After rigid-body refinement of the whole search model and the five fragments the initial electron density map revealed the location of the MgADP and the AlF4- moiety. Further refinement with simulated annealing reduced the R-factor to 31% (Briinger, 1990b). Additional conventional least squares refinement revealed the location of the NH2-terminal domain (Briinger, 1990b; Tronrud et al., 1987). There are breaks in the polypeptide chain between Gln 19 and Leu 27, Ala 205 and Gly 209, Phe 487 and Leu 508, Arg 620 and Asn 626, and Glu 681 and Gly 683, where the density is poorly defined. A total of 646 residues was included in the present model together with 104 water molecules and the nucleotide complex. The final R-factor between the observed and calculated data was 24.2% for all data between 30 and 2.6 resolution where the rms deviation from ideality for the bond lengths, angles, and trigonal planes was 0.013 A, 2.39", and 0.016 A, respectively. Most of the residues between Asp 2 and Lys 690 are well defined; however, the COOH-terminal 70 residues appear to adopt multiple conformations in the crystal and have not been modeled at present. In both structures,

RESULTS AND DISCUSSION

Tertiary Structure of MgADPeBeFx*SI Dc. The tertiary structure of the beryllium fluoride complex is described first because it most closely matches that of chicken skeletal muscle myosin S 1 and because it is a more complete model. The overall topology of the protein will be described with respect to the three major tryptic fragments of skeletal muscle myosin S1, the NH2-terminal 25 kDa, the central 50 kDa, and the COOH-terminal20 kDa fragment (Baht et al., 1975; Cooke, 1986; Mornet et al., 1979).

A ribbon drawing of the MgADP*BeFx*SIDc complex is shown in Figure 3, where the molecule is oriented to view into the narrow cleft that splits the central 50 kDa segment of the myosin heavy chain. A comparison of the chicken and DictyosteZium myosin heads is shown in Figure 4a. The overall domain structure and topology of the truncated head of DictyosteZium myosin is remarkably similar to that of the chicken skeletal muscle myosin S1 with which it shares 47% sequence identity (Warrick et al., 1986). The rms difference

8964 Biochemistry, Vol. 34, No. 28, 1995 Fisher et al.

1986) and thus serves to confirm both the chicken and Dictyostelium chain tracings. The structural similarity seen here proves that reductive methylation found necessary to obtain crystals of chicken S1 (Rayment et al., 1993b) did not alter the domain structure of the myosin head. It is noteworthy that the narrow cleft that splits the 50 kDa segment of the heavy chain is equally open in both chicken and MgADPBeFyS 1Dc structures. Contrary to expectations, the nucleotide binding pocket in MgADPBeFx*S 1 Dc has only closed about 1 A relative to that observed in the chicken skeletal myosin S1.

In the MgADPAlF4-S 1 Dc complex, the tertiary structures of the individual domains that form the NHz-terminal, upper and lower central segments and COOH-terminal segments of the heavy chain are very similar to those of the MgADPBeFyS 1Dc complex (Table 2). However, the relationship between the domains has changed (Figure 4b). The most obvious difference is the partial closure of the cleft that splits the central segment of the myosin heavy chain. This results in a significant rearrangement of the COOH- terminal segment of the polypeptide chain beyond Lys 690. The reason for this conformational change originates in difference in the coordination of the beryllium and aluminum fluoride moieties in the phosphate binding pocket. The final segment of the model includes Cys 678 and Thr 688 that are equivalent to the reactive cysteine residues (SH2 and SH1) in skeletal muscle. Even though Cys 678 and Thr 688 are still located at opposite ends of an a-helix as observed in chicken myosin subfragment-1 (Rayment et ai., 1993b), the orientation of the helix has changed by -20'. The significance of this movement is discussed briefly later. However, due to the disorder present in the MgADPAIF4-S 1Dc complex the discussion of these structures will focus on the catalytic mechanism for ATP hydrolysis.

Structure of the Adenine and Ribose Binding Pocket. The nucleotide binding site lies at the base of a pocket formed on one side by residues from the NH2-terminal segment of the myosin heavy chain and on the other by residues from the central section as indicated in Figure 3 and shown in detail in Figure 5. There is well defined electron density in both the MgADPBeF,*S 1Dc and MgADP-AlF4.S 1 Dc com- plexes for ADP, the metallofluoride complex, a magnesium ion, and several water molecules in the nucleotide binding pocket (Figure lb,c). The coordination of the adenine and ribose is essentially identical in both complexes. Conse- quently, the discussion of this component of the nucleotide binding site will be described relative to the structure of MgADPBeF,-S 1Dc since this is the better model.

In both complexes the purine ring is in the anti conforma- tion relative to the ribose, which exhibits a C3'-endo ring pucker. Almost all of the residues involved in the binding of the nucleotide lie above the plane of the p-sheet that forms the major tertiary structural motif in the myosin head. However, the pocket extends down into the opening between the cleft that splits the upper and lower domains of the central 50 kDa segment. A schematic drawing of the myosin- nucleotide interactions in the MgADPBeF,*S 1 Dc complex is shown in Figure 6. The binding site for the purine ring in both complexes is formed almost exclusively from residues in the NHz-terminal segment of the chain. It lies in a small indentation formed on one side by the first turn of the helix that follows the characteristic phosphate binding-loop se- quence (Walker et al., 1982) and involves residues Gly 184,

a S1 dc-MgADP*BeFx

-1.1 X l x . . ...................... ......x+....,

-180 -120 -60 . '%, I

0 60 120 180 -180

b 180

120

60

3 0

-60

-120

-1%

Q

S1 dc-MgADP.AIF4'

. . ..........................................

X

X '

X X

X X

X

X

X

60 im Q

FIGURE 2: Ramachandran plot of the main-chain dihedral angles for (a) MgADPBeFx*SIDc and (b) MgADPAlF461Dc. All non- glycinal dihedral angles are indicated as dots whereas the glycinal angles are shown as crosses. Fully allowed 4 and 1/, values are enclosed by continuous lines; those only partially allowed are enclosed by broken lines.

between 554 equivalent a-carbons in chicken and Dictyo- stelium myosin is 1.25 A. This is only slightly larger than the expected difference of 1.1 A for proteins that share the same function and exhibit 47% sequence identity (Chothia & Lesk, 1986). A comparison of the rms differences between both of the metallofluoride-ADP complexes of SlDc and chicken skeletal muscle myosin S1 is given in Table 2. The NH2- and COOH-tenninal regions of the Dictyostelium myosin sequence were excluded from the initial refinement and were built independently of the known fold for chicken myosin S1. Although there is limited sequence identity in these regions, the similarity in topology is consistent with the conserved function (Chothia & Lesk,

Structural Studies on Myosin-Nucleotide Complexes Biochemistry, Vol. 34, No. 28, 1995 8965

FIGURE 3: Ribbon representation of the MgADP*BeFx*SIDc complex of the truncated Dictyostelium myosin. The heavy chain is displayed in green, red, and blue to delineate th minal, central, and COOH-terminal fragments that extend from residues Asn 2 to Ala 205, Gly 209 to Ala 62 1, and Phe 627 to respectively. These segments are separated by loops for which the electron density is ill- defined. In addition to t ort loops that connect these segments, residues Thr 501 to Gly 507 are disordered. This figure and figures 4, 5, and 7 were prepared the molecular graphics program MOLSCRIPT (Kraulis, 199 1).

a

b I I I

I I I

FIGURE 4: Stereo a-carbon plots of the alignment of the motor units of (a) chicken skeletal muscle myosin and the MgADPOBeF *SlDc and (b) MgADPBeF,*S 1 Dc and MgADP*AlF& 1 Dc complexes of Dictyostelium n in green, red, and blue to distinguish the NH2-termina1, central, and COOH-terminal s structures of chicken skeletal muscle myosin S1 and MgADP.AIF&lDc are shown in blac

over the purine ring, protecting it from externa forms a salt bridge with Glu 187. In many myosins Arg 131 and Glu 187 are a try an and valine, respectively. The equivalent tryptophan r ue in skeletal muscle myosin

The other side of the indentation is hat forms the top of one side of the

ocket and includes contributions from Asn 127, Pro 128, Arg 13 1, and Tyr 135. Arg 131 folds

8966 Biochemistry, Vol. 34, No. 28,1995 Fisher et al.

Table 2: rms Differences (A) between Structurally Equivalent a-Carbon Positions in Chicken Skeletal Myosin Subfragment- 1 and the MgADPBeFx-S 1Dc and MgADP*AlF& 1Dc Complexes"

N-terminal upper 50 kDa lower 50 kDa C-terminal whole molecule domain domain domain domain

MgADPBeFx*S 1Dc onto chicken S 1 0.73 (1 17) 1.20 (225) 0.89(118) 1.15 (109) 1.25 (554)

MgADPBeFx*S 1 Dc onto MgADPAlF4.S 1 Dc 0.62 (186) 0.63 (244) 0.91 (132) 0.99 (54) 1.51 (640)

a The numbers of structurally equivalent a-carbon atoms used in the superposition calculation are given in parentheses. Surface loops which showed large deviations in a-carbon positions were omitted from the calculations. The structures have been divided into four parts representing the NH2-terminal domain, the upper and lower domains of the central 50 kDa segment, and the COOH-terminal domain. In the whole molecule superpositions between the chicken structure and the two Dict-yostelium structures, the first 84 residues of the N-terminus were omitted due to large differences in the trace of the polypeptide chains. In addition, the last 60 residues of the C-terminal region have also been omitted in the chicken- MgADPBeFx*SIDc superposition, as the polypeptide chains are in a slightly different location even though they have an almost identical fold.

MgADPAlF4S 1 Dc onto chicken S 1 0.95 (1 17) 1.25 (225) 1.33 (116) 1.42 (49) 1.91 (504)

r ,

1 , I.'

aspect of nucleotide binding. In the structure of chicken skeletal myosin this residue faces away from the active site pocket, whereas in Dictyostelium myosin this residue forms a salt bridge with Asp 113, which also lies outside the nucleotide binding pocket. The latter residue is not highly

ross the myosin superfamily, which suggests e is not a conserved feature of the myosin most skeletal muscle myosins examined Lys

thylated. In many other myosins the equivalent

has been identifie

three additional met this time it is not cl

is less stable than introduction of a hy

tion of the neighboring residues that form the purine binding site.

The ribose moiety of ADP forms very few specific interactions with the protein or water molecules. However, Asn 127, which is highly conserved, does form a hydrogen bond to the ring oxygen, 04'. The 2'- and 3'-hydroxyl groups are both exposed to solvent, an arrangement that accounts for the observation that a wide variety of functional

Structural Studies on Myosin - Nucleotide Complexes Biochemistry, Vol. 34, No. 28,1995 8967

FIGURE 6: Schematic representation of the coordination of the metallofluoride- ADP complex in MgADPBeF$l Dc. Hydrogen bonds less than 3.5 8, and ionic interactions less than 3.7 8, are shown as dashed lines with the corresponding distances. Only those interactions with the appropriate stereochemistry to participate in these interactions are included. In this figure beryllium, carbon, fluorine, magnesium, nitrogen, oxygen, and phosphorus are depicted in green, black, yellow, orange, blue, red, and pink, respectively.

groups including spin-labels, fluorophores, and photolabels can be attached to the ribose ring in these positions without affecting function (Cole & Yount, 1990; Cremo et al., 1990; Crowder & Cooke, 1987). This accounts for the ability of myosin to utilize organotriphosphates that replace the ribose ring with an aminoethyl group and the purine ring with derivatives of nitrobenzene (Wang et al., 1993).

Structure of the Triphosphate Binding Pocket. The coordination of the a- and @-phosphates is very similar in both MgADPBeF,*S 1 Dc and MgADPAlF& 1Dc. However, there are significant differences between the coordination of the beryllium and aluminum fluoride moieties. The coor- dination of the a- and @-phosphate and the metallofluoride moiety in the MgADPBeF,*S 1Dc complex is described first since this mimics the state of the molecule prior to ATP hydrolysis. In the MgADPBeF,*S 1Dc complex three well- defined peaks of electron density are associated with a terminal oxygen atom of the @-phosphate (Figure lb). These have been attributed to three fluorine atoms bonded to a beryllium atom. There is little electron density associated with the beryllium since this atom only carries two electrons. Although the coordination surrounding the central beryllium atom is well defined, this study is unable to distinguish between a fluorine or hydroxyl ligand. However, the coordination is consistent with three fluorine ligands. Spec- troscopic studies of the beryllium fluoride- ADP adduct with smooth muscle myosin suggest that this complex is a mixture of fluoroberyllates (Henry et al., 1993). The geometry of the heavier atoms and the predicted bond lengths are consistent with an approximately tetrahedral BeF30 adduct. Furthermore, the predicted beryllium to @-oxygen distance of 1.57 A is indistinguishable from the bridging oxygen distances between the a- and @-phosphate groups (1.56 and 1.67 A, respectively). Likewise the Be-F bond distances are similar to the nonbridging P-0 bond distances. The MgADPBeF, complex is thus an analog of ATP in the active site.

In the MgADPBeF,*S 1 Dc complex the a- and @-phosphate groups and metallofluoride moiety extend down into a narrow tunnel formed by the phosphate binding loop Gly 179 to Val 187 on one side and residues Asn 233 to Gly 240 on the other. This tunnel opens into the apex of the cleft that splits the upper and lower domains of the central 50 kDa segment of the heavy chain. The magnesium ion is located at the top of the tunnel (Figure 5) and is surrounded by six ligands in an approximately octahedral arrangement and plays a central role in chelating the nucleotide complex where it interacts with both the P-phosphate and the BeF, moiety. The Mg2+ ion is also coordinated to two protein ligands, Oy 1 of Thr 186 and Oy of Ser 237 and two water molecules. These two water molecules are an integral part of the coordination scheme of the nucleotide. One of these water molecules interacts with the carboxyl side chain of Asp 454 and also forms a bond to a fluorine atom of BeF,. The second interacts with the main-chain carbonyl oxygen of Asn 235 and one of the @-phosphate oxygen atoms.

There are an extensive series of specific interactions between the a- and @-phosphate groups and BeF, moiety with the protein (Figure 6). Every oxygen and fluorine of the pseudo-triphosphate moiety is involved in a hydrogen- bonding interaction. However, there are more interactions to the @-phosphate and BeF, moiety than the a-phosphate group. The phosphate binding loop contributes six hydrogen bonds from its main-chain amide nitrogens and additional interactions from the side chains of Ser 181, Thr 186, and Lys 187. The opposite side of the binding pocket formed by residues Asn 233 to Gly 240 contributes six hydrogen bonds from both the backbone and side chains to the @-phosphate and BeF, moiety either directly or via a buried water molecule. Oy of Ser 236 is hydrogen-bonded to one of the fluorine atoms. The analogous atom in skeletal muscle myosin was identified previously to be a ligand of the y-phosphate from photoreaction of the vanadate- ADP complex (Grammer & Yount, 1991).

8968 Biochemistry, Vol. 34, No. 28,1995 Fisher et al.

Comparison of the Metallofluoride Complexes. Super- position of 'four regions (Table 2) of MgADPBeFx*SIDc structure onto the corresponding regions of MgADPAlF& 1Dc and chicken myosin S 1 indicates that the overall fold of these parts of the S 1 molecule is essentially identical in all of these structures. The rms difference between the M and MgADPBeFx*S 1Dc structures is 1.5 BiADFAlFdSlDC which results from the partial closure of the narrow cleft in the central 50 kDa segment (Figure 7a). Analysis of the MgADPAlF&lDc model superimposed onto MgADPBeFx*S lDc, matching the first part of the central 50 kDa segment, Gly 209 to Asp 454, indicates that the closure results from a concerted rigid- body movement of the lower domain, Gly 457 to Ser 619; the amount of closure is estimated to be about 5". The overall rms difference between the MgADPAlFsS 1 Dc and MgADPBeFx*S 1Dc underestimates the significance of this domain movement, since the outer edge of the cleft moves by more than 5 A.

The movement is essentially a bending at residues Ile 455 and Gly 457, located at the end of a /?-strand which drops from the upper part of the central domain into the lower. The bending at Ile 455 involves a single rotation about the main-chain conformational torsion angle 4 (the torsion angles 4 and ly, are -152" and 155" in MgADPBeFx*SIDc, -155" and 169" in chicken S1, and -59" and 166" in MgADP AlF4*SlDc), while at Gly 457, the change involves the ly angle (-70" and 136" in MgADPAlF4.SlDc compared to -89" and -162" in MgADPBeFx*SIDc). The conforma- tional torsion angles in residues on either side of this hinge region remain virtually unchanged in all structures; the rms difference in 4 and ?$ angles for five amino acids on either side of Ile 455 (Ile 464 in chicken S1) is *20".

The flexing at Ile 455 and Gly 457 brings the loop from Ser 456 to Ser 465 and the NHz-terminal region of the following helix (Phe 466 to Cys 470) into close proximity to a region in the upper central domain near the active site. Five new interdomain contacts are introduced: a hydrogen bond from the 0, of Ser 456 to the carbonyl oxygen of Ser 237, a salt bridge between the side chains of Glu 459 and Arg 238, and hydrogen bonds between the side chains of Glu 467 (OC2) and Ser 266 (Oy), and Gln 468 (Nf2) and Glu 264 (Ofl), and between the carbonyl oxygen of Ile 460 and the side chain of Arg 232. In addition, there are water- mediated interactions between the side chains of Ser 181, Thr 230, and Arg 238 (N,2) in the upper central domain and Glu 459 ( 0 , 2 ) in the lower. Finally, there is the direct interaction between the main-chain nitrogen of Gly 457 and one of the fluoride ions of the AlF4- moiety.

Movement of the lower domain relative to the upper domain of the central segment of the heavy chain in the MgADPAIF&lDc complex is coupled to a structural change in the COOH-terminal domains. These are mediated by the loop extending from Tyr 573 to Gly 575 that abuts the final helix observed in the MgADP.AlF4-S 1 Dc complex. This helix separates Cys 678 and Thr 688 that are the equivalent residues to the reactive cysteine residues (SH2 and SH1) in skeletal muscle. A 5 8, movement of Tyr 573 induces an -20" rotation and 2.5 8, translation of that helix relative to its prior location (the change in the disposition of this helix is evident in Figure 4b). Unfortunately, the polypeptide chain beyond Lys 690 in the MgADPAlF4.S 1 Dc complex appears to adopt multiple conformations in the crystal. This lack of conformational stability is most likely due to the absence of

The bottom of the nucleotide binding pocket abuts the apex of the narrow cleft that splits the 50 kDa segment of the heavy chain. The region around the metallofluoride moiety will be referred to as the y-phosphate pocket. This pocket contains at least three well-ordered water molecules. One of these forms a short hydrogen bond, 2.5 A, to one of the fluorine atoms. This water molecule is roughly opposite the phosphate oxygen that forms the pseudobridge to beryllium. It is expected that hydrolysis of ATP occurs through nucleophilic attack of the y-phosphorus by a water molecule approaching from a direction opposite to the bridging oxygen. The water molecule observed in the MgADPBeFx*S 1Dc complex is well situated for this purpose. The other water molecules in the y-phosphate pocket participate in stabilizing the infrastructure surrounding the beryllium fluoride moiety.

The overall coordination of the magnesium ion and a- and /?-phosphate groups in MgADPAlF4-S 1Dc is very similar to that seen in the MgADPBeFx*SIDc complex (Figure 7). Only five ligands are observed for the magnesium ion, whereas six were seen in the beryllium complex. The water molecule that bridges the magnesium ion and Asp 454 in MgADPBeFx*SIDc is not evident in MgADPAlF4.S lDc, even though the disposition of all of the other ligands is identical in both complexes. This difference is most likely due to the lower resolution and less complete refinement in the MgADPAlF4-SlDc complex. The location of the aluminum in the structure of the MgADPAlF4-S 1Dc complex is unambiguous (Figure IC). Although the arrangement of ligands is less well defined, due to the resolution of the data, the overall coordination of the metal ion is octahedral. One of the ligands is contributed by an oxygen atom of the /?-phosphate. Four of the remaining ligands lie in a plane and are most likely fluorine atoms, which is consistent with spectroscopic studies of this complex in smooth muscle myosin (Maruta et al., 1993). The electron density for the water molecule, axial to the pseudobridging oxygen atom, is not well defined at the present resolution and has not been included in the model. This water molecule would be analogous to that coordinated to the BeFx moiety close to the axial position (Figure lb). The stereochemistry observed here for the AlF4- moiety is similar to that seen in the structures of the G-protein Gial and transducin a complexes with guanosine diphosphate (GDP) and aluminum fluoride (Coleman et al., 1994; Sondek et al., 1994). There are several other well-ordered water molecules in the y-phos- phate pocket in locations analogous to those seen in the MgADPBeFx*SIDc structure.

The bond distance to the pseudobridging oxygen is distinctly longer (2.0 A) in the MgADPAlF4.SlDc complex than in the MgADPBeFx*S 1Dc (1.57 A). This change allows one of the fluorine ligands to the aluminum to form a short hydrogen bond to the amide hydrogen of Gly 457, which stabilizes a movement of the main chain by 4.4 A. Gly 457 is completely conserved in the myosin superfamily. In the MgADPBeF,*S 1Dc complex the respective fluorine is too far away to form this interaction. Because of the longer bond distance to pseudobridging oxygen observed in the MgADP AlF4.S 1Dc complex it appears that this complex is an analog of the transition state in comparison to MgADPBeFx*SIDc which more closely mimics the prehydrolysis state (ATP). The MgGDPAlF4 structures of the G-protein Gial and transducin a complexes were also interpreted as transition state analogs (Coleman et al., 1994; Sondek et al., 1994).

Structural Studies on Myosin-Nucleotide Complexes

I a

Biochemistry, Vol. 34, No. 28, I995 8969

b

FIGURE 7: Structural differences in the y-phosphate pockets of the MgADP-BeFL*S 1Dc and MgADP*AlF4*S 1Dc complexes. (a) Tertiary rences between the MgADPBeFx*Sl Dc (blue) and MgADPAlF4-S 1Dc (red) complexes. The additional interactions between lower domains of the central segment of the myosin heavy chain prompted by the AlF4- ion are included. Side-chain and tions in MgADPAlF&lDc are shown in gray and their corresponding orientation in MgADPBeFx*SIDc in green. (b) and ews of the interactions of the P-phosphate, metallofluoride ions, and water molecules in the phosphate binding pocket in

MgADPBeFx*S 1Dc and MgADP-AlF& 1 Dc, respectively.

the essential light chain. so suggests that a different explain the cross-linking of the reactive cysteine residues in structural change in the terminal domain would be skeletal muscle in the presence of ATP (Huston et al., 1988), observed if the essenti chain was present. As a although they d icate that this segment of the structure consequence the current structures are inappropriate to undergoes a c ational change during hydrolysis.

8970 Biochemistry, Vol. 34, No. 28,1995

Mechanism of Hydrolysis. Kinetic studies have shown that ATP binding is a multistep process that is followed by rapid hydrolysis, which results in a metastable complex of myosin, ADP, and inorganic phosphate (Pi) (Lymn & Taylor, 1971; Trybus & Taylor, 1982). The equilibrium constant between ATP and ADP,Pi is approximately unity on the enzyme (depending on the ionic strength and pH) (Bagshaw & Trentham, 1973, 1974; Lymn & Taylor, 1971). During this time there is rapid interconversion between substrate and products (Bagshaw & Trentham, 1973; Trentham et al., 1976). Hydrolysis of ATP occurs by attack of a water molecule on the y-phosphorus.

Although myosin has been studied extensively, very little is known about the role of the protein in ATP hydrolysis. In particular the identity of the catalytic base, if it exists, has not been defined. Examination of the y-phosphate pocket in both MgADPBeF,*S 1 Dc and MgADPAlF4.S 1 Dc does not reveal any amino acid side chains within 5 A of the beryllium or aluminum that might function as a catalytic base (Figure 7c). The closest groups to the aluminum are Oy of Ser 236 (4.0 A), Oy of Ser 237 (3.8 A), Oy of Ser 181 (4.0 A), and NC of Lys 185 (3.8 A) (Figure 7b). Additional groups in the y-phosphate pocket include Glu 459 and Ser 456. Ser 18 1 lies in the plane of the AlF4- ion and is hydrogen-bonded to one of the fluorine ligands. Ser 236 is hydrogen-bonded to one of the fluorine ligands. Ser 237 is coordinated to the magnesium ion and thus cannot function as the base. Likewise Lys 185 is coordinated to both a P-phosphate oxygen and one of the fluorines and would not be expected to function as a base. Glu 459, which is hydrogen-bonded to a water molecule in the y-phosphate pocket, is too far from the aluminum to function as a base. If a water molecule were included in the coordination sphere for the aluminum, it would be located 3.4 8, from Ser 456. Oy of Ser 456 is 5.2 from the aluminum and directly opposite the pseudo- bridging oxygen of the P-phosphate. There are at least four serine residues associated with the y-phosphate pocket. However, unless one of these is activated, as in serine proteases, it would not be expected to function as a base. Examination of the structure does not reveal any side chains that might serve to reduce the pKa of these serine residues. However, all of these residues are completely conserved.

The lack of an obvious candidate for a base suggests that water might function as the nucleophile with direct transfer of its proton to the y-phosphate. This would eliminate the need for an amino acid side chain to function as the base and would allow the protein to catalyze the reaction by stabilizing a trigonal-planar arrangement of oxygen ligands around the y-phosphorus atom at the transition state. A similar mechanism has been suggested previously for trans- ducin a (Sondek et al., 1994) and p2lraS (Schweins et al., 1995) Altematively, one of the serine residues associated with the y-phosphate pocket might serve as an intermediate in the transfer of the proton from the nucleophilic water molecule to the y-phosphate (Figure 8). In this altemative mechanism, the serine would participate in hydrogen ex- change but would not function as a catalytic base. The rate constant for the bond cleavage step in ATP hydrolysis by myosin is approximately 5 s-l at 3 "C (Taylor, 1977) and is smaller than the minimum rate of hydrogen exchange of 30 s-l for serine hydroxyl protons as measured by NMR at 4 "C (Liepinsh et al., 1992). The latter occurs in neutral solution at pH 6.5 in model peptides; however, the exchange

Fisher et al.

dr 236 d r 236 FIGURE 8: Proposed mechanism of hydrolysis for myosin where Ser 236 acts as an intermediary in the proton transfer from the hydrolytic water molecule to the y-phosphate.

rate would be considerably higher in the polar environment of the y-phosphate pocket. A possible candidate for this type of proton transfer is Ser 236. In both complexes, Oy of Ser 236 is coordinated to one of the fluorine atoms and is situated appropriately for interaction with the anticipated incoming water molecule. This mechanism would provide better stereochemistry for proton transfer from the incoming water to the y-phosphate than direct transfer.

Myosin is unusual in that it forms a metastable state (ADP,Pi) after hydrolysis. In the absence of actin, product release is slow. Kinetic studies have shown that there is no release of protons during the bond cleavage (Bagshaw & Trentham, 1973) and that this occurs at the time of product release. Furthermore, I8O exchange studies have suggested that the phosphate group of the product is tumbling rapidly in the active site since all of the nonbridge phosphate oxygen atoms are equivalent (Bowater et al., 1990; Trentham et al., 1976). The cavity associated with the putative location of the y-phosphate is large enough to accommodate a phosphate group and contains sufficient hydrogen bond donors and acceptors to coordinate this ion. In the MgADPAlF4.S 1Dc complex, which mimics the transition state for hydrolysis, there is reduced access of solvent to the y-phosphate pocket caused by partial closure of the cleft. If this conformational change is maintained in the metastable ADP,Pi state, it provides a structural explanation for the kinetic stability this state.

Kinetic studies indicate that the phosphate ion is released prior to ADP during the contractile cycle (Goldman, 1987; Trentham et al., 1976). The arrangement of MgADP and the metallofluoride in both structures suggests the phosphate ion cannot leave via the same route that the substrate enters. It is unlikely that the phosphate ion could squeeze past MgADP. The kinetic data and X-ray structures indicate that phosphate may depart from the bottom of the active site pocket as has been suggested from modeling ATP in the active site of myosin (Yount et al., 1995).

The original model for the contractile cycle derived from the structure of chicken skeletal muscle myosin S1 suggested that the nucleotide pocket closed when ATP bound to generate a bent molecule (Rayment et al., 1993a). An assumption was made that the conformation of chicken skeletal myosin S 1 observed in that structure was intermedi- ate between the actin and ATP-bound state of the molecule. The structure of the MgADPBeF,*S 1Dc complex indicates that the motor unit of chicken skeletal myosin S1 when methylated and crystallized from ammonium sulfate is essentially identical to the ATP-bound state. There are several ways to interpret this observation. First, the con- formation of myosin S1 in solution might be closer to the nucleotide-bound state. This would then account for the very

Structural Studies on Myosin-Nucleotide Complexes

high binding constant of myosin for ATP and would explain the difficulty of detecting any major macroscopic change in its structure upon addition of ATP with most physical techniques. Alternatively, the structure of the chicken S1 might be a function of the methylation or the high concentra- tion of ammonium sulfate that is known to be a competitive inhibitor of ATP binding (Tesi et al., 1988). Indeed, a sulfate ion was observed in the nucleotide pocket of the chicken S 1 structure (Rayment et al., 1993b). It is difficult to separate these possibilities. However, it would appear from these structural studies that closure of the nucleotide pocket is not the source of the major conformational change that is responsible for the molecular basis of muscle contraction. This is consistent with recent fluorescence quenching measurements, which show that the purine base of adenine experiences a similar environment at the beginning and end of the contractile cycle (Franks-Skiba et al., 1994). The structures of the beryllium and aluminum fluoride complexes suggest that movements in the myosin heavy chain associated with the lower domain of the central 50 kDa segment coupled to movements in the COOH-terminal segment are the primary source of t h i s conformational change. The implica- tions of this modification to the structural model for the contractile cycle have been discussed elsewhere (Fisher et al., 1995).

CONCLUSION The structure of the truncated myosin head from Dicryu-

stelium is remarkably similar to that of chicken skeletal myosin S 1 considering that both light chains are missing from the former structure. The differences between the NH2- and COOH-terminal segments of MgADPBeFx*S 1Dc and MgADP AIF4.S 1 Dc reported here re-emphasize the importance of the essential light chain in energy transduction (Lowey et al., 1993). It has been suggested that the light chain binding motif serves to amplify the conformational changes in the head and increase the size of the power stroke (Rayment et al., 1993b). This has been substantiated by genetic experi- ments that removed the binding site for the regulatory light chain (Uyeda & Spudich, 1993). However, it now appears that the interactions between the essential light chain and the motor unit are important for transmitting the conforma- tional changes induced by ATP binding and hydrolysis to the remainder of the molecule. This is consistent with the observation that a significant fraction of mutations in B-cardiac myosin that have been implicated in familial hypertrophic cardiomyopathy map to the interface between the essential light chain and the heavy chain (Rayment et al., 1995).

The current structural results emphasize the importance of the narrow cleft that splits the central 50 kDa segment of the heavy chain. They further suggest that it not only functions in sensing the presence of the y-phosphate of ATP but also is responsible for transducing the conformational change that results in the power stroke. The new information derived from the structure of the truncated Dicryostelium head shifts the location of the conformational change in the structural hypothesis from the nucleotide binding pocket to the region associated with the reactive cysteine residues. This is consistent with the chemical evidence that implicates conformational changes in this region during the hydrolytic process (Burke & Reisler, 1977; Dalbey et al., 1983; Wells & Yount, 1980). These new structural results provide further

Biochemistry, Vol. 34, No. 28,1995 8971

evidence that actomyosin-based motility is driven by con- formational rearrangement of domains.

ACKNOWLEDGMENT

We thank G. H. Reed for helpful discussions and W. W. Cleland for suggesting the potential role of serine in the hydrolytic mechanism. We thank Siemens (Madison) for the loan of a HI-STAR double detector system used in this study.

REFERENCES Bagshaw, C. R., & Trentham, D. R. (1973) Biochem. J . 133,323-

Bagshaw, C. R., & Trentham, D. R. (1974) Biochem. J. 141,331-

Balint, M., Sreter, F. A,, Wolf, I., Nagy, B., & Gergley, J. (1975)

Bivin, D. B., Ue, K., Khoroshev, M., & Morales, M. F. (1994)

Bowater, R., Zimmerman, R. W., & Webb, M. R. (1990) J. Biol.

Briinger, A. T. (1990a) Acta Crystallogr. A46, 46-57. Briinger, A. T. (1990b) X-PLOR Version 3.1: A System for

Crystallography and NMR, Yale University, New Haven, CT. Burke, M., & Reisler, E. (1977) Biochemistry 16, 5559-5563. Chothia, C., & Lesk, A. M. (1986) EMBO J. 5, 823-826. Cole, D., & Yount, R. G. (1990) J. Biol. Chem. 265,22547-22546. Coleman, D. E., Berghuis, A. M., Lee, E., Linder, M. E., Gilman,

Cooke, R. (1986) CRC Crit. Rev. Biochem. 21, 53-1 18. Cremo, C. R., Neuron, J. M., & Yount, R. G. (1990) Biochemistry

Crowder, M. C., & Cooke, R. (1987) Biophys. J. 51, 323-333. Dalbey, R. E., Weiel, J., & Yount, R. G. (1983) Biochemistry 22,

4696-4706. Fisher, A. J., Smith, C. A., Thoden, J., Smith, R., Sutoh, K., Holden,

H. M., & Rayment, I. (1995) Biophys. J. 68, 19s-28s. Franks-Skiba, K., Hwang, T., & Cooke, R. (1994) Biochemistry

33, 12720-12728. Goldman, Y. E. (1987) Ann. Rev. Physiol. 49, 637-654. Goodno, C. C. (1982) in Methods in Enzymology (Frederiksen &

Cunningham, Eds.) pp 116-123, Academic Press, New York. Grammer, J. C., & Yount, R. G. (1991) Biophys. J. 59, 226a. Henry, G. D., Maruta, S., Ikebe, M., & Sykes, B. D. (1993)

Biochemistry 32, 1045 1 - 10456. Holmes, K. C., Popp, D., Gebhard, W., & Kabsch, W. (1990)

Nature 347, 44-49. Huston, E. E., Grammer, J. C., & Yount, R. G. (1988) Biochemistry

27, 8945-8952. Huxley, A. F., & Niedergerke, R. (1954) Nature 173, 971-973. Huxley, H., & Hanson, J. (1954) Nature 173, 973-976. Itakura, S., Yamakawa, H., Toyoshima, Y. Y., Ishijima, A,, Kojima,

T., Harada, Y., Yamnagida, T., Wakabayashi, T., & Sutoh, K. (1993) Biochem. Biophys. Res. Commun. 196, 1504-1510.

Jones, T. A. (1985) in Methods in Enzymology (Wycoff, H . W., Hirs, C. H. W., & Timasheff, S. N., Eds.) pp 157-171, Academic Press Inc., New York.

Kabsch, W., Mannherz, H. G., Suck, D., Pai, E. F., & Holmes, K. C. (1990) Nature 347, 37-44.

Kraulis, P. J. (1991) J. Appl. Crystaflogr. 24, 946-950. Liepinsh, E., Otting, G., & Wlithrich, K. (1992) J. Biomol. NMR

Lowey, S., Waller, G. S., & Trybus, K. M. (1993) Nature 365,

Lundblad (1991) Chemical Reagents for Protein Modi'jZcation, 2nd

Lymn, R. W., & Taylor, E. W. (1971) Biochemistry 10, 4617-

Manstein, D. J., Titus, M. A., De-Lozanne, A., & Spudich, J. A.

Mamta, S., Henry, G. D., Sykes, B. D., & Ikebe, M. (1993) J. Biol.

Means, G. E., & Feeney, R. E. (1968) Biochemistry 7,2192-2201.

328.

349.

J . Biol. Chem. 2.50, 6168-6177.

Proc. Natl. Acad. Sci. USA. 91, 8665-8669.

Chem. 265, 171-176.

A. G., & Sprang, S. R. (1994) Science 265, 1405-1412.

29, 3309-3319.

2,447-465.

454-456.

ed., CRD Press, Boca Raton, FL.

4624.

(1989) EMBO J . 8,923-932.

Chem. 268, 7093-7100.

8912 Biochemistry, Vol. 34, No. 28,1995

Milligan, R. A., & Flicker, P. F. (1987) J . Cell Biol. 105, 29-39. Milligan, R. A., Whitaker, M., & Safer, D. (1990) Nature 348,217-

Momet, D., Pantel, P., Audemard, E., & Kassab, R. (1979) Biochem.

Muller, C. W., & Schulz, G. E. (1992) J . Mol. Biol. 224, 159-

Navaza, J. (1993) Acta Crystallogr. 0 4 9 , 588-591. Pai, E. F., Krengel, U., Petsko, G. A., Goody, R. S., Kabsch, W.,

Pate, E., Franks-Skiba, K., White, H., & Cooke, R. (1993) J . Biol.

Phan, B., & Reisler, E. (1992) Biochemistry 31, 4787-4793. Phan, B. C., Faller, L. D., & Reisler, E. (1993) Biochemistry 32,

7712-77 19. Phan, B. C., Cheung, P., Miller, C. J., Reisler, E., & Muhlrad, A.

(1994) Biochemstry 33, 11286-11295. Rayment, I., Holden, H. M., Whittaker, M., Yohn, C. B., Lorenz,

M., Holmes, K. C., & Milligan, R. A. (1993a) Science 261,58- 65.

Rayment, I., Rypniewski, W. R., Schmidt-Base, K., Smith, R., Tomchick, D. R., Benning, M. M., Winkelmann, D. A., We- senberg, G., & Holden, H. M. (1993b) Science 261, 50-58.

Rayment, I., Holden, H. M., Sellers, J. R., Fananapazir, L., & Epstein, N. D. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 3864- 3868.

221.

Biophys. Res. Commun. 89, 925-932.

177.

& Wittinghofer, A. (1990) EMBO J . 9 , 2351-2359.

Chem. 268, 10046-10053.

Read, R. J. (1986) Acta Crystallogr. A42, 140- 149. Rossmann, M. G. (1972) The Molecular Replacement Method,

Ruppel, K. M., Uyeda, T. Q. P., & Spudich, J. A. (1994) J . Biol.

Rypniewski, W. R., Holden, H. M., & Rayment, I. (1993)

Schweins, T., Geyer, M., Scheffzek, K., Warshel, A., Kalbitzer,

Smith, T. J. (1993) J . Appl. Crystallogr. 26, 496-498. Sondek, J., Lambright, D. G., Noel, J. P., Hamm, H. E., & Sigler,

Gordon and Breach, New York.

Chem. 269, 18773-18780.

Biochemistry 32, 9851 -9858.

H. R., & Wittinghofer, A. (1995) Struct. Biol. 2, 36-44.

P. B. (1994) Nature 372, 276-279.

Fisher et al.

Taylor, E. W. (1977) Biochemistry 16, 732-740. Tesi, C., Barman, T., & Travers, F. (1988) FEBS Lett. 236, 256-

Tong, L., & Rossmann, M. G. (1990) Acta Crystallogr. A46,783-

Tong, S. W., & Elzinga, M. (1983) J . Biol. Chem. 258, 13100-

Trentham, D. R., Eccleston, J. F., & Bagshaw, C. R. (1976) Q.

Tronrud, D. E., Ten Eyck, L. F., & Matthews, B. W. (1987) Acta

Trybus, K. M., & Taylor, E. W. (1982) Biochemistry 21, 1284-

Uyeda, T. Q. P., & Spudich, J. A. (1993) Science 262, 1867-1870. Walker, J. E., Saraste, M., Runswick, M. J., & Gay, N. J. (1982)

Wang, D., Pate, E., Cooke, R., & Yount, R. (1993) J . Muscle Res.

Warrick, H. M., De Lozanne, A,, Leinwand, L. A,, & Spudich, J.

Wells, J. A., & Yount, R. G. (1980)Biochemistry19, 1711-1717. Werber, M. M., Peyser, M., & Muhlrad, A. (1992) Biochemistry

White, H. D., Belknap, B., & Jiang, W. (1993) J . Biol. Chem. 268,

White, H. W., & Rayment, I. (1993) Biochemistry 32,9859-9865. Xie, X., Harrison, D. H., Schlichting, I., Sweet, R. M., Kalabokis,

V. N., Szentz-Gyorgyi, A. G., & Cohen, C. (1994) Nature 368, 306-312.

Yount, R. G., Cremo, C. R., Crammer, J. C., & Kerwin, B. A. (1992) Philos. Trans. R. Soc. London B336, 55-61.

Yount, R. G., Lawson, J. D., & Rayment, I. (1995) Biophys. J . (in press).

BI950582A

260.

792.

13110.

Rev. Biophys. 9 , 217-281.

Crystallogr. A43, 489-501.

1294.

EMBO J . 1, 945-951.

Cell Motil. 14, 484-497.

A. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 9433-9437.

31, 7190-7197.

10039- 10045.